Lineage tracking discloses a contribution of endothelial cells to hematopoiesis in adult BM.

(A) Experimental design: tamoxifen was administered to 8- to12-week-old Cdh5-Cre mice to induce fluorescent labeling of VE-Cadherin⁺ cells and their cell progeny. Four weeks later, BM and blood were analyzed. (B) CD31⁺EGFP+ BM ECs in Cre⁻ mice (n=10) and Cre⁺ mice treated with oil (n=13) or tamoxifen (n=10); flow cytometry results. (C and D) CD45+EGFP+ cells in BM and blood from Cre⁻ mice (n=8) and Cre⁺ mice treated with oil (n=6-10) or tamoxifen (n=15-18). Representative flow cytometry gating in Figure S1G. (E) Representative blood smear from a tamoxifen-treated Cdh5-CreERT2(PAC)/ZsGreen mouse showing ZsGreen⁺CD45⁺DAPI⁺ cells (arrows). (F) Kinetics of ZsGreen⁺ cell detection in BM ECs (CD45⁻VE-Cadherin⁺) and blood white blood cells (WBC) post-tamoxifen; mouse n=8-10/group). (G) EGFP+ B and T-lymphocytes, granulocytes, and monocytes in BM of tamoxifen-treated mice (n=14) as percent of total EGFP+ cells; 3 experiments. (H) EGFP⁺ BM LSK, lymphocytes, granulocytes, and monocytes as percent of total EGFP⁺/- cell type; Cdh5-CreERT2(PAC)/mTmG mice (oil n=10; tamoxifen n=15), 3 experiments. (I) UMAP plots of Lin⁻ BM HSPC from tamoxifen-treated Cdh5-CreERT2(PAC)/ZsGreen mice (n=26; 1 femur/mouse) showing FlowSOM clustering of all (ZsGreen⁺/-) and ZsGreen⁺ populations. (J) Violin plots showing ZsGreen⁺ cell distribution across HSPC subsets from (I). Dots represent individual mice; data shown as mean±SD except shown as median in (G). *p < 0.05, **p < 0.01, ***p < 0.001 by Student’s t test.

BM ECs generate engraftable hematopoietic cells ex vivo.

(A) BM cells from tamoxifen-treated mice were cultured on high-attachment Primaria flasks or OP9 cell monolayers. Representative images show ZsGreen⁺ cells at weeks 1, 3, and 8. (B) Workflow for culturing unsorted and sorted BM cell populations. Post-sort purity of ZsGreen+ ECs is shown in the bottom left panel. All cells were cultured (8 weeks) on OP9 cell monolayers supplemented with WT BM cells. Culture medium and floating cells were removed twice/week for 7 weeks. At the start of week 8, one final WT BM and medium supplementation was implemented prior to harvest at the end of week 8. (C and D) Representative flow cytometry plots (C) and quantification (D) of CD45⁺ZsGreen⁺ cells from each of the 8-week cultures (n=5). (E) Floating/loosely adherent ZsGreen⁺ cells from unsorted BM 8-week cell cultures were sorted and transplanted (5×104, 2.5×104,1.25×104 or 6.25×103 cells) into lethally irradiated (11 Gy) WT mice (n=2/group). (F and G) Representative flow cytometry image (F) and quantification (G) of low-adherent cells harvested after 8 weeks of culture, showing that >95% (group average) of ZsGreen⁺ low-adherent cells are CD45⁺. These ZsGreen+CD45+ cells were sorted for transplantation. (H - J) WBC counts from 5 control mice (no irradiation or transplant) (H) and percent ZsGreen⁺, ZsGreen dim and ZsGreen⁻ cells (I-J) in blood of transplant recipients 10 months post-transplant (n=6). Dots represent individual mice. Data are shown as mean ± SD. ns, not significant by Student’s t test.

Adult BM endothelial cells give rise to hematopoietic cells following transfer into conditioned recipients.

(A) Transplant experiment: donor ECs from BM of tamoxifen-treated mice were FACS-sorted and transplanted into WT C57Bl/6 recipients conditioned with 5-FU or PBS. (B) ZsGreen⁺ ECs detected in BM of 5-FU-conditioned (n=5) or PBS-conditioned (n=15) recipients of ECs 4 weeks post-transplant. (C and D) ZsGreen+CD45+ cells (C) and cell type distribution (D) in the BM and blood of 5-FU-conditioned transplant recipients of BM ECs or no cell controls (n=5/group). (E and F) Age-dependent decline of ZsGreen⁺CD45⁺ cells (E) but not ZsGreen+VE-Cadhenin+ cells (F) in the BM of Cdh5-CreERT2(BAC)/ZsGreen mice (n=35) treated with tamoxifen 4-weeks prior to harvest. (G and H) Cell number (G; mouse n=8-12) and cell type distribution (H; mouse n=6) in the peritoneal cavity (PerC) of PBS- or thioglycolate (TGL)-pretreated (4 hours) mice. (I) ZsGreen⁺ and ZsGreen⁻ PerC cell types in TGL-pretreated mice (n=12). (J) Representative histograms depicting pHrodo Red fluorescence detection of E-Coli phagocytosis. (K and L) E. coli⁺ phagocytosis by ZsGreen⁺ and ZsGreen⁻ PerC neutrophils (K) and macrophages (L) in TGL-pretreated mice (n=4). (M) Representative histograms depicting CellRox Orange fluorescence for cell-associated ROS detection. (N and O) CellRox mean fluorescence intensity (MFI) in ZsGreen⁺ and ZsGreen⁻ PerC neutrophils (N) and macrophages (O) in TGL-pretreated mice (n=4). Dots represent individual mice. Data are shown as mean ± SD. *p < 0.05, ***p < 0.001, ns, not significant by Student’s t test.

Independence of adult EHT from preexisting HSPC.

(A) Transplantation experiment: donor LSK sorting, recipient irradiation, transplantation, tamoxifen treatment, and analysis (top). Tabular representation of possible outcomes of the experiment designed to address the question “Do HSPCs/other hematopoietic cells in Cdh5CreERT2/ZsGreen mice express Cdh5-CreERT2?” (bottom). (B-D) Blood WBC counts (B), percent ZsGreen⁺ PBMC (C), and time course of ZsGreen⁺ PBMC detection (D) in transplant recipients of ZsGreen⁻ LSK (5×104 or 2.5×104 cells/mouse; n=3/group) and ZsGreen-enriched LSKs (2.8×103 cells/mouse; n=2). Results in B and C are from week 24 post-tamoxifen. (E) Experiment: WT BM transplantation (BMTP) into lethally irradiated Cdh5-Cre/mTmG mice (n=9). Four weeks later, tamoxifen was administered; blood was monitored for 16 weeks. (F and G) EGFP⁺ PBMC detection before and after tamoxifen or peanut oil administration (F) and cell type distribution of EGFP⁺ and EGFP⁻ PBMCs at week 12 post-tamoxifen or peanut oil (G) in Cdh5-Cre/mTmG recipients (n=9) of WT BM (5×106 cells). Statistical significance reflects comparisons between EGFP+ and EGFP- cells in the tamoxifen vs peanut oil groups. Dots represent individual mice. Data are shown as mean ± SD. ***p < 0.001, ns, not significant by Student’s t test.

Polylox sc lineage tracing links adult BM ECs to hematopoietic progenitors and mature blood cell progeny.

(A) Schematic of Polylox barcode and transcriptome profiling. FACS-enriched ECs (ZsGreen⁺VE-Cadherin+Endomucin+) and EC-depleted (ZsGreen-VE-Cadherin-Endomucin-) BM cells from tamoxifen-treated Cdh5-CreERT2/ZsGreen/PolyloxExpress mice (n=3, 10-week-old at the time of tamoxifen treatment) were mixed (1:1), and encapsulated (147,446 cells loaded; 93,553 processed). Indexed cDNA was used for scRNA-seq and barcode detection by PacBio sequencing after nested PCR enrichment; barcode-transcriptome integration was accomplished via shared cell indices. (B) UMAP clustering and cell type annotation. Clusters 0, 1, 13, and 22 comprise ECs; cluster 14 comprises Mesenchymal-type cells. (C – E) Heatmaps showing “true” Polylox barcodes (pGen < 1×10⁻6) linking HSPCs to hematopoietic cells (C), ECs to hematopoietic and other cells (D), and Mesenchymal-type cells to other cells (E). The numbers within the colored boxes identify cell number; the labels at the bottom of each column denote the barcode shared by all cells in that column; the number on the right side the heatmaps reflects the total number of cells in each row. (F) UpSet plot showing cells (identified by colored dots) sharing the same “true” barcode (identified by lines connecting the colored dots); bar graph at the top of the plot reflects (height and number on each bar) the number of “true” barcodes. Colors of dots: EC (red), Mesenchymal-type (orange), ECs connecting with Mesenchymal-type cells (blue), cells other than ECs and Mesenchymal-type cells (black). (G) Violin plots showing selected gene expression profile in Mesenchymal-type cells (cluster 14) and ECs (clusters 0, 1, 13, 22 combined).

Sc transcriptomic analysis of prospective hemogenic ECs.

(A) UMAP clustering of 434,810 cells from eight public scRNA-seq datasets. (B) Dot plot showing relative Cdh5 and Runx1 co-expression across clusters; clusters 8 and 50 co-express both genes. (C) UMAP highlighting clusters 8 and 50; all other clusters shown in grey. (D) Violin plots of doublet scores across Leiden clusters. Clusters 50 and 8 show no evidence of doublet enrichment. (E) Datasets proportional contribution to clusters 50 and 8; each dataset is color-coded. (F) Dot plot showing expression of selected marker genes in clusters 50 and 8 (from the public sc RNA-seq datasets listed in Figure 7D) and from clusters 0, 1, 13, 22 and 14 (from Polylox scRNA-seq; Figure 5B). Results reflect mean expression and fraction of cells in group. (G) Cdh5, Runx1 and Col1a2 co-expression in the indicated clusters as a fraction of cells in the cluster. (H and I) t-SNE plot of ECs from 11 murine tissues (G) and Venn diagram (H) showing rare co-expression of Cdh5, Runx1, and Col1a2 in these tissues.

Contribution of Col1a2 and Runx1 expression to ECs hemogenic activity.

(A and B) Percent EGFP+CD45+ cells in BM and blood of tamoxifen-treated (n=6) or oil-treated (n=5) Col1a2-CreERT2/mTmG mice (A) and tamoxifen-treated (n=4) or oil-treated (n=3) Col1a2-CreERT2/ZsGreen mice (B). Cre-control mice (n=5 in A, and n=2 in B). (C) Transplant experiment: sorted VE-Cadherin⁺CD45⁻ZsGreen⁺/Col1a2⁺ cells from tamoxifen-treated Col1a2-CreERT2/ZsGreen mice are transplanted into 5-FU-conditioned WT recipients. (D and E) Detection (D) and characterization (E) of ZsGreen⁺CD45⁺ cells in BM and blood of WT 5-FU-conditioned mice (n=5), 4 weeks post-transplant of VE-Cadherin⁺CD45⁻ZsGreen⁺/Col1a2⁺ cells. Control FU-conditioned WT mice (n=4) received no cell transplant (D). (F) Time course of ZsGreen⁺ PBMC detection in control (Cdh5-Cre⁺/ZsGreen⁺) and Runx1EC-KI (Cdh5-Cre⁺/ZsGreen⁺/Runx1-KI) mice (n=10 per group). (G and H) Representative images (G) and quantification (H) of ZsGreen⁺ cells from OP9 cell-supported cultures of BM cells from tamoxifen-treated Cdh5-Cre⁺/ZsGreen⁺ (n=11) and Runx1EC-KI mice (n=5). (I and J) Representative flow cytometry plots (I) and quantification (J) of CD45⁺ZsGreen⁺ cells from OP9 cell-supported BM cell cultures (n=5/group). Dots represent individual mice. Data are shown as mean ± SD. **p<0.01, ***p < 0.001 by Student’s t test.

Contribution of ECs to hematopoiesis in adult BM is revealed by Cdh5-CreERT2 mouse tracking lines.

Related to Figure 1. (A) Cdh5-tracking mouse lines. Tamoxifen switches-on green fluorescence in cells that express the Cre-recombinase and their cell progeny. (B and C) Confocal microscopy images of representative BM sections from Cdh5-CreERT2(PAC)/ZsGreen (B) and Cdh5-CreERT2(BAC)/ZsGreen (C) adult mice showing tamoxifen-induced ZsGreen fluorescence co-staining of most Endomucin+ cells. Control BM sections from representative Cre+ mouse treated with peanut oil (no tamoxifen) display occasional ZsGreen+Endomucin+ cells but no tamoxifen-independent ZsGreen fluorescence is detected in representative Cre- mice. (D) Flow cytometry analysis of adult BM cells from Cdh5-CreERT2(PAC)/ZsGreen (n = 6-8) and Cdh5-CreERT2(BAC)/ZsGreen mice (n = 10) shows that ZsGreen fluorescence identifies most ECs four weeks after tamoxifen administration but also tracks a small proportion of EC expressing tamoxifen-independent fluorescence in Cre+ but not Cre- mice. (E) Percent CD45+ZsGreen+ cells of viable BM cells from Cre- control (n=6), Cre+ control (peanut-oil treated, no tamoxifen; n= 8) and Cre+ tamoxifen-treated Cdh5-CreERT2(PAC) /ZsGreen (n=6) or Cdh5-CreERT2(BAC)/ZsGreen mice (n=9). Mice were 8 to 12 weeks old at the time of tamoxifen administration. (F) Percent ZsGreen+ cells of PBMC from Cre- control, Cre+ control (peanut-oil treated) and Cre+ tamoxifen-treated Cdh5-CreERT2(PAC)/ZsGreen and Cdh5-CreERT2(BAC)/ZsGreen mice (n=9/group). Mice were 8 to 12 weeks old at the time of tamoxifen administration. (G–H) Representative flow cytometry gating of CD45⁺EGFP⁺ cells from BM and blood of Cdh5-CreERT2/mTmG mice (G, relates to Figure 1 C, D), and CD45⁺ZsGreen⁺ cells from BM and blood of Cdh5-CreERT2/ZsGreen mice (H, relates to Figure S1 E, F). (I) Representative confocal images showing a nucleated (DAPI+) BM ZsGreen+CD45+ cell in the BM from a Cdh5-CreERT2(PAC)/ZsGreen mouse treated with tamoxifen. (J) Percent ZsGreen+ cells of PBMC in individual Cdh5-CreERT2(BAC)/ZsGreen mice before or four weeks after tamoxifen administration. Each dot represents the results from 50-250µl blood/mouse. The lines link results from individual mice (n=15, 8-12 weeks old). (K) Percent B and T-lymphocytes, granulocytes, and monocytes among EGFP+ PBMC of Cre+ tamoxifen-treated Cdh5-CreERT2(PAC)/mTmG mice (n=15, 8-12 weeks old). (L) Percent EGFP+ cells in peripheral blood cell populations of Cre+ peanut oil-treated (n=9) and tamoxifen-treated (n=16) Cdh5-CreERT2(PAC)/mTmG mice (8-12 weeks old). Dots represent individual mice. Data are shown as mean ± SD. *p<0.05, **p<0.01, ***p < 0.001 by Student’s t test.

Characterization of tracked hematopoietic progenitors and mature cells in adult BM and peripheral blood of Cdh5-Cre reporter mice.

Related to Figure 1, 2, or 3. (A and B) Percent B lymphocytes, T lymphocytes, granulocytes, and monocytes of all ZsGreen+ cells (A) and percent ZsGreen+ cells of total BM LSK, B and T-lymphocytes, granulocytes, and monocytes (B). Each dot reflects results from individual mice (1 femur plus 1 tibia combined, n=11); group means ± SD (error bars) are shown by the horizontal lines. (C and D) Percent B lymphocytes, T lymphocytes, granulocytes, and monocytes of all ZsGreen+ PBMC in Cdh5-CreERT2(PAC)/ZsGreen mice treated with tamoxifen (C, n=10) and percent ZsGreen+ cells of peripheral blood B lymphocytes, T lymphocytes, granulocytes, and monocytes in Cdh5-CreERT2(PAC)/ZsGreen mice (D) treated with peanut oil (n=15) or tamoxifen (n=20). (E) Gating strategy for identification of HSPC subsets in bone marrow. Representative flow cytometry plots show sequential gating of lineage negative (Lin-) Sca1+ cKit+ (LSK) cells into long-term hematopoietic stem cells (LT-HSC), short-term hematopoietic stem cells (ST-HSC), multipotent progenitors (MPP), common lymphoid progenitors (CLP), common myeloid progenitors (CMP), megakaryocyte-erythroid progenitors (MEP), and granulocyte-macrophage progenitors (GMP). (F) Representative cytospin image of floating and low-adherent cells from ex vivo culture of BM cells from a tamoxifen-induced Cdh5-CreERT2(PAC)/ZsGreen mouse. (G) Representative image (relates to Figure 2B) showing the appearance of sorted ZsGreen⁺ ECs after 4-week culture on OP9 monolayer. (H) Gating strategy for sorting VE-Cadherin+ ZsGreen+ ECs from the BM of Cdh5-CreERT2(PAC)/ZsGreen mice. (I) Purity analysis of sorted VE-Cadherin+ ZsGreen+ ECs. (J) Gating strategy used for detecting ZsGreen+CD45+ hematopoietic cells in WT C57Bl/6 recipients of ECs sorted from the BM of Cdh5-CreERT2(PAC)/ZsGreen donors. (K) Percent CD45+ZsGreen+ cells recovered from the peritoneal cavity (PerC) of Cre+Cdh5-CreERT2/ZsGreen mice treated with peanut oil (n=8) or tamoxifen (n=15). Cre- mice (n=5). (L) Percent ZsGreen+ cells within PerC cell populations recovered from mice (n=15) under steady-state conditions. Cell type identification; B1 cells: CD19+, CD3-, CD45R(B220)-, CD5+, CD43+; B2 cells: CD19+, CD3-, CD45R(B220)+, CD5-, CD43-; T cells: CD3+, CD11b-; monocytes: CD11b+, CD19-, Ly6G-, Ly6Chigh; neutrophils: CD11b+, CD19-, Ly6G+, Ly6Clow; and macrophages: CD11b+, CD19-, F4/80+. Dots represent individual mice. Data are shown as mean ± SD. *p<0.05, **p<0.01, ***p < 0.001 by Student’s t test.

BM pDCs, but not other BM hematopoietic cells, express Cdh5 and Ptprc, encoding CD45.

Related to Figure 4. (A) UMAP plot showing unsupervised clustering of sc transcriptomic data from a public dataset of mouse BM hematopoietic cells. (B and C) UMAP plots displaying expression of Cdh5 (B) and Ptprc (encoding CD45, C) in the dataset shown in (A). (D) Dot plot illustrating the expression of selected marker genes across clusters shown in (A). The red rectangle highlights gene expression by cluster 13 cells, identifying pDCs. Dot size represents the percentage of cells expressing the gene within each cluster, and color intensity reflects the mean expression level. (E) Gating strategy for selecting ZsGreen- and ZsGreen+ LSK progenitors. (F and G) Analysis of purity of sorted LSK populations enriched for ZsGreen+ cells (F) and depleted of ZsGreen+ cells (G) from the BM of Cdh5-CreERT2(PAC)/ZsGreen mice (not treated with tamoxifen).

Single-cell RNA-seq analysis of BM ZsGreen+ cells from tamoxifen-treated Cdh5-Cre/ZsGreen/Polylox mice.

Related to Figure 5. (A) UMAP plot showing unsupervised clustering of sc RNA-seq data, identifying 34 distinct cell clusters within BM ZsGreen+ cells. (B) Histogram of doublet score distribution. A threshold of 0.2 was applied to match the expected doublet rate from 10x Genomics Chromium GEM-X chips. (C) Doublet score distribution (gray violin plots, left Y-axis) and corresponding doublet percentages (blue bars, right Y-axis) across Leiden clusters. (D) UMAP plots of clusters after doublet removal, and expression of ZsGreen1, Cdh5, Pecam1, Eng, CreERT2 and Ptprc (CD45). (E) Dot plot showing expression of selected marker genes across Leiden clusters identified in (A). Dot size indicates the proportion of cells expressing the gene; color intensity reflects the average expression level of each cluster.

Identification and distribution of ‘True’ Polylox barcodes across cell types.

Related to Figure 5. (A) Bubble plot showing individual Polylox barcodes plotted against their corresponding pGen values (log₁₀ scale). The y-axis indicates representative barcodes (one label is shown for every five barcodes). Bubble size reflects the number of cells harboring each barcode. The red dashed line denotes the log₁₀(pGen) = −6 cutoff, which was used to define true barcodes retained for downstream analyses. (B) UMAP plot showing the distribution of ‘True’ Polylox barcodes across the 34 Leiden-defined clusters identified in Figure S4A. Cells containing ‘True’ barcodes (n = 721) are shown in orange; cells with ‘Not True’ barcodes (n = 3,348) are shown in blue. A total of 388 barcodes were detected, including 274 ‘True’ (orange) and 125 ‘Not True’ (blue), as listed below the UMAP plot. (C) Heatmap displaying the distribution and abundance of ‘True’ Polylox barcodes across annotated cell types. Each row corresponds to a unique barcode (1 out of every 5 barcodes shown); color intensity represents the number of cells carrying that barcode within each listed cell type. (D) Heatmap showing cell cycle phase distribution (G1, G2/M, S) across Leiden clusters identified in Figure S4A. Color intensity and numerical values represent the percentage of cells in each phase within the indicated cell type.

Characterization of Col1a2-tracked cell populations in BM and blood.

Related to Figure 7. (A) Schematic representation of the Col1a2 tracking lines. (B) Representative confocal image of a BM section from a tamoxifen-treated Col1a2-CreERT2/ZsGreen mouse, showing widespread distribution of ZsGreen⁺ cells. (C to E) Flow cytometric identification of RUNX1+VE-Cadherin+CD45- ECs in the BM of peanut oil-treated ( n=6) and tamoxifen-treated (n=6) Col1a2-CreERT2/ZsGreen adult mice; Cre-mice (n=5) (C); WT C57Bl/6 mice (n= 6) and Fluorescence Minus One (FMO) control (n=5) (D); and Cdh5-CreERT2(PAC)/ZsGreen mice treated with peanut oil (n=6) or tamoxifen (n=6); Cre- mice (n=5)(E). Left panels: quantification of cells identified by the indicated gates as a percentage of total BM ECs; each dot represents one mouse (1 femur + 1 tibia). Middle and right panels: representative gating strategies. (F) Representative confocal microscopy image of a BM section from a tamoxifen-treated Col1a2-CreERT2/ZsGreen adult mouse showing a ZsGreen+ Endomucin+ cell lining a vascular structure (white arrows). (G) Representative confocal image of a blood smear from a Col1a2-CreERT2/ZsGreen mouse treated with tamoxifen showing the presence of a nucleate CD45+ cell tracked by ZsGreen/Col1a2 fluorescence (pointed by the arrow). (H) Representative confocal image of a blood smear from a Col1a2-CreERT2/mTmG mouse treated with tamoxifen showing the presence of a nucleated CD45+ cell tracked by EGFP/Col1a2 fluorescence (pointed by the arrow). Dots represent individual mice. Data are shown as mean ± SD. ***p < 0.001 by Student’s t test.

Analysis and hemogenic potential of Col1a2-tracked adult BM ECs.

Related to Figure 7. (A) Representative FACS gating strategy used to isolate BM hematopoietic cells, ECs, stromal cells, and Col1a2-tracked ECs from tamoxifen-treated Col1a2-CreERT2/ZsGreen mice. (B) Gene expression profiling of unsorted BM and sorted BM populations defined in (A). Results from qRT-PCR are normalized by Gapdh and unsorted BM. Dots reflect experimental triplicates. (C) Representative confocal image of a BM section from a transplant recipient showing a CD45⁺CD11b⁺ZsGreen⁺ cell (arrow), indicating hematopoietic derivation from transplanted ZsGreen⁺VE-Cadherin⁺CD45⁻ (Col1a2⁺) cells. (D) Representative image of a blood smear from a WT recipient mouse transplanted with ZsGreen (Col1a2)+ BM ECs from Col1a2-CreERT2/ZsGreen mice. ZsGreen tracked cells are pointed by the arrows. (E) Schematic diagram of the Cdh5-CreERT2/ZsGreen/Runx1-Knock-in (Runx1EC-KI) mouse line used to trace endothelial cells with Runx1 expression induced upon tamoxifen treatment.